Greetings!
when i use apigee x reverse proxy to call a target url of another apigee edge gateway, and under concurrent load of 20 same users per 3 seconds using jmeter, i noticed a strange behavior, which is as follows:
1. all requests are sent to the target and the target responded with 200 OK.
2. Apigee x only receives around 90% of the responses with 200 OK and the rest returns 504, 502.
3. each time i get the same exact behavior with slightly different error rate.
4. when i try same use case from jmeter but calling edge directly all requests are OK and respond in max 2 seconds / avg 700 ms.
5. the apigee x reverse proxy is empty with no added policies/sharedflows
6. when i test apigee x with isolation (mocking by calling aws eks deployed wiremock ) all requests are OK.
i have great confidence that the network calls from apigee x reverse proxy to edge have these intermittent errors only during response. (some responses are lost on the middle of the way back to apigee x)
so,
I'd like to know if there is some explanation to this and potential reasons / solutions.
also, if i can implement retries on apigee x so that when the response delay happens i set max delay to be 2 sec for example and the retry instead of waiting 55 (edge)/ 60 (X) seconds until either timeout then retry.
by the way, I'm not using target servers , i just use direct url in apigee x
LinkedIn
Twitter
